A structural study on the Alzheimer’s disease amyloid β peptide

A structural study on the Alzheimer’s disease amyloid β peptide

Date: June 12, 2008

Name: Maria Koster

Student number: 1223626

School: Hogeschool Utrecht, Life Sciences & Chemistry

Major: Molecular Biology

Internship location: Texas A&M University

Department of Biochemistry & Biophysics

College Station, Texas, USA

Period: September 1^st 2007 till May 31^st 2008 Supervisors: Dr. J. C. Sacchettini

J. Mire

Mentor: Dr. M. F. van Berlo

Preface

From September 2007 till May 2008 I have done an internship in the Sacchettini Lab in the Department Biochemistry and Biophysics of Texas A&M University in College Station. There are no words for how much I have learned and experienced during this period in my life, on a professional as well as personal level. I can truly say that it has been an amazing and unforgettable experience.

Of course this internship would not have been possible without the help and effort of others. I would like to thank Dr. James Sacchettini for giving me the opportunity to work in his lab. I would like to thank Joseph Mire for teaching and guiding me, thank you so much! Further I would like to thank Dr. Mario van Berlo for helping me to find this internship place. Last but not least I want to thank all the people in the lab. Everybody has been so helpful and nice to me; it made me feel like being at home. I will never forget you guys. I wish you all the best in life!

Maria Koster

College Station, May 2008

Summary

Alzheimer’s disease (AD) is a neurodegenerative disease. The disease is characterized by the presence of neuropathology hallmarks such as: senile plaques, neurofibrillary tangles and the loss of synapses and neurons in the brain. The senile plaques are mainly composed of amyloid fibrils of the protein amyloid β (Aβ)^4,5.

Aβ arises naturally in the human brain throughout life as a result of metabolic processing of the amyloid precursor protein (APP)¹². There are several Aβ peptides, which originate after cleavage of APP. The 42 residue long peptide, Aβ42 is the major component of the senile plaques in AD¹⁵. Alzheimer’s disease is thought to develop through a process called amyloid cascade where an increase of Aβ level results in oligomerization and aggregation of Aβ42 thereby triggering a downstream cascade of pathological events³. More and more evidence indicates that especially the soluble Aβ oligomers (that are assemblies of the Aβ monomer) are neurotoxic and play an important role in the AD related pathology^4,5,23,25.

The ultimate aim of this study was to solve the structure of Aβ42 in an oligomeric state for rational structure based drug design against Alzheimer’s disease. Protein purification and X-ray crystallography techniques were used. Aβ was linked to the apical domain of the chaperonin protein GroEL and the fusion protein ApicalGroEL-Aβ was created. This fusion inhibited the rapid fibrillization of Aβ, thereby forming the possibility to trap Aβ in an oligomeric state. Eleven different constructs were made. They contained a N- terminal His-tag followed by the apical domain of GroEL. In between the apical domain and Aβ different synthetic sequences, the so-called linkers were placed to manipulate the arrangement of Aβ. Also, different size combinations of the apical domain of GroEL and Aβ were tested.

The ApicalGroEL-Aβ constructs were expressed and purified and two different crystallization approaches were used. The first approach was to isolate the protein as a monomer and screen for crystallization. Since Aβ has the propensity to form oligomers when it is at a high enough concentration¹⁷, the assumption was made that the protein would oligomerize in the crystal. The second approach was to isolate ApicalGroEL-Aβ in an oligomeric state and to crystallize the protein in this oligomeric state. The protein monomer was purified, brought to high concentration and incubated to see if it would form oligomers in a time dependent manner, which could then be purified and crystallized.

Structures of three different ApicalGroEL-Aβ constructs in a monomeric state were solved. The electron density maps showed electron density for the apical domain of GroEL, up to the last residue. No connected and continuous electron density was present for the linker and Aβ. Further analysis suggested that Aβ is present in the crystals. The missing electron density could have been due to too much flexibility of Aβ in the unit cell. It could also be due to a disordered conformation of Aβ itself.

The ApicalGroEL-Aβ monomer oligomerizes in a time dependent manner. No attempt was made yet to isolate and crystallize these oligomers. The best approach would be to

isolate and crystallize the smaller size oligomers (dimers and trimers) since these are stable^23,24.

Linking Aβ to the apical domain of GroEL offers advantages; the ApicalGroEL-Aβ protein is soluble and easy to purify and crystallize. In addition the apical domain of GroEL itself does not oligomerize, therefore a stable oligomer formed is due to the presence of Aβ.

Unfortunately it was not possible to solve an oligomer structure of Aβ yet. However there are still multiple approaches that can be tried, hopefully one of them will eventually enable us to solve the structure of Aβ in an oligomeric state. Thereby making it possible to design drugs that specifically target Aβ oligomers and help to win the fight against Alzheimer’s disease.

Samenvatting

De ziekte van Alzheimer (AD) is een neurodegeneratieve ziekte. De ziekte wordt gekarakteriseerd door de aanwezigheid van neuropathologische kenmerken waaronder:

plaques, neurofibrillaire knopen en de afname van synapsen en neuronen in de hersenen.

De plaques bestaan voornamelijk uit amyloïde fibrillen van het eiwit amyloid β (Aβ)^4,5. Aβ komt van nature voor in de humane hersenen tijdens het leven en ontstaat door metabole verwerking van het amyloïd voorloper eiwit (APP)¹². Er zijn verscheidene Aβ peptides die ontstaan na klieving van APP. De 42 residu lange peptide, Aβ42 is de voornaamste component in de plaques in AD¹⁵. De ziekte van Alzheimer ontwikkelt zich waarschijnlijk via de zogenaamde amyloïde cascade waarbij een toename van Aβ leidt tot oligomerizatie en aggregatie van Aβ42 wat een cascade van pathologische gebeurtenissen tot gevolg heeft³.

Steeds meer bewijs wijst er op dat voornamelijk de oplosbare Aβ oligomeren (dat zijn verzamelingen van Aβ monomeren) neurotoxisch zijn en een belangrijke rol spelen in de AD gerelateerde pathologie^4,5,23,25.

Het doel van deze studie is het bepalen van de structuur van Aβ42 in een oligomerische vorm, voor op structuur gebaseerde medicijnontwikkeling gericht tegen de ziekte van Alzheimer. Technieken zoals eiwitzuivering en röntgenstraling crystallografie werden toegepast. Aβ werd gelinkt aan het apicale domein van het chaperonine eiwit GroEL waarbij het fusie-eiwit ApicalGroEL-Aβ werd gecreëerd. Deze fusie voorkomt snelle fibrillizatie van Aβ, hierdoor wordt de mogelijkheid gevormd om Aβ te vangen in een oligomerische vorm. Elf verschillende constructen zijn gemaakt. Ze hadden een N- terminale His-tag gevolgd door het apicale domein van GroEL. Tussen het apicale domein van GroEL en Aβ zijn verschillende synthetische sequenties, de zogenaamde linkers geplaatst om de ordening van Aβ te manipuleren. Daarnaast zijn verschillende combinaties in grootte van het apicale domein van GroEL en Aβ uitgetest.

De ApicalGroEL-Aβ constructen werden tot expressie gebracht en gezuiverd. Vervolgens zijn twee verschillende kristallisatiebenaderingen toegepast. Aβ heeft de neiging om oligomeren te vormen als de concentratie hoog genoeg is¹⁷, daarom werd aangenomen dat het eiwit zou oligomerizeren in de kristal. De tweede benadering was het isoleren en kristalliseren van ApicalGroEL-Aβ in een oligomerische vorm. De eiwit monomeer werd gezuiverd en tot hoge concentratie gebracht. Vervolgens werd het geïncubeerd over een periode om te zien of het oligomeren zou vormen die gezuiverd en gekristalliseerd konden worden.

De structuur van drie verschillende ApicalGroEL-Aβ constructen in een monomerische vorm zijn bepaald. De 3D afbeelding van de electronendichtheid liet electronendichtheid voor het apicale domein van GroEL zien tot en met de laatste residu. Geen aaneengesloten en continue electronendichtheid was aanwezig voor de linker en Aβ. De niet aanwezige electronendichtheid kan het gevolg zijn van een te grote flexibiliteit van Aβ in de eenheidscel. Het kan ook een gevolg zijn van dat Aβ zich in een ongeordene conformatie bevond.

De ApicalGroEL-Aβ monomeer oligomeriseert na een incubatieperiode. Het is nog niet

gepoogd om deze oligomeren te isoleren en kristalliseren. De beste benadering zou het isoleren en kristalliseren van de kleinere oligomeren (dimeren en trimeren) zijn aangezien deze stabiel zijn^23,24.

Het linken van Aβ aan het apicale domein van GroEL biedt voordelen; het ApicalGroEL- Aβ eiwit is oplosbaar en gemakkelijk te zuiveren en kristalliseren. Ook oligomeriseert het apicale domein van GroEL niet uit zichzelf, elk stabiel gevormde oligomeer is daarom afhankelijk van de aanwezigheid van Aβ.

Helaas was het niet mogelijk om een oligomeerstructuur van Aβ te bepalen. Er zijn nog verschillende benaderingen die uitgeprobeerd kunnen worden, hopelijk zal een van deze het mogelijk maken op de structuur van Aβ in een oligomerische vorm te bepalen.

Daarmee zou de mogelijk gecreëerd worden om medicijnen te ontwikkelen die zich specifiek richten op Aβ oligomeren. Dit kan helpen om het gevecht tegen de ziekte van Alzheimer te winnen.

Abbreviations

Å Ångstrom

Aβ Amyloid beta

AD Alzheimer’s disease

AICD APP intracellular domain

APP Amyloid precursor protein

AU Absorbance unit

BACE β-site APP cleavage enzyme

CTF C-terminal fragment

DLS Dynamic light scattering

DS Down Syndrome

FPLC Fast protein liquid chromatography

GAG Glycosaminoglycan

IPTG Isopropyl-β-thiogalactopyranoside

LB Luria-Bertani

LTP Long-term potentiation

mQ MilliQ water

Mw Molecular weight

NCT Nicastrin

NFT Neurofibrillary tangle

NTF N-terminal fragment

O.D. Optical density

PDB ID Protein Data Bank identification

PMSF Phenylmethanesulfonyl fluoride

PS Presenilin

qHX-NMR Quenched hydrogen-exchange nuclear magnetic resonance ssNMR Solid state nuclear magnetic resonance

U Unit

PREFACE...3

SUMMARY ...4

SAMENVATTING...6

ABBREVIATIONS ...8

CHAPTER 1; INTRODUCTION ...12

1.1 BIOCHEMICAL BACKGROUND OF AMYLOID BETA...12

1.1.1 Amyloid diseases...12

1.1.2 Alzheimer’s disease...12

1.1.3 Amyloid beta discovery and formation process...13

1.1.4 Amyloid beta fibrillization process ...14

1.1.5 Amyloid beta oligomer pathology...15

1.1.6 Amyloid beta oligomer kinetics ...16

1.1.7 Common fibril and oligomer structures of amyloids ...17

1.1.8 Amyloid beta proposed structures ...18

Conformation transition from α-helical to β-sheet...18

Amyloid beta monomer α-helical structures in apolar solution...19

Amyloid beta monomer collapsed coil structure in polar solution...19

Amyloid beta fibril β-sheet structures ...19

1.1.9 Drugs against Alzheimer’s disease...20

1.2 PROTEIN CRYSTALLOGRAPHY...21

1.2.1 Crystallization ...21

1.2.2 X-ray diffraction ...22

1.2.3 The phase problem...23

1.2.4 Model building...23

1.3 AIM AND PROCEDURE...24

CHAPTER 2; MATERIALS AND METHODS...27

2.1 GENERAL EXPERIMENTS: PLASMID, GENE AND PROTEIN CONSTRUCTS...27

2.2 GENERAL EXPERIMENTS: CLONING OF THE APICALGROEL-Aβ CONSTRUCTS...28

2.2.1 Plasmid preparation ...28

Double digestion...28

DNA extraction from agarose gel...28

2.2.2 Insert preparation ...29

Linker preparation ...29

Gene preparation...29

2.2.3 Ligation and transformation...30

Linker ligation into cut plasmid ...30

Gene ligation into cut plasmid...30

Plasmid transformation into NovaBlue competent cells...31

2.2.4 ApicalGroEL-Aβ construct confirmation...31

Overnight cultures ...31

Plasmid isolation...31

Sequencing to confirm construct ...32

2.3 GENERAL EXPERIMENTS: EXPRESSION OF THE APICALGROEL-Aβ CONSTRUCTS...32

2.3.1 Plasmid transformation into BL21 competent cells...32

2.3.2 Expression test ...32

2.4 GENERAL EXPERIMENTS: APICALGROEL-Aβ PURIFICATION...33

2.4.1 Cell growth and protein production ...33

2.4.2 Cell lysis ...33

2.4.3 Nickel affinity chromatography ...33

2.5 APICALGROEL-Aβ ISOLATION AS A MONOMER...34

2.5.1 Crystallization preparation ...34

2.5.2 Crystallization screening...34

2.5.3 Crystallization optimization ...35

2.5.4 Structure determination and refinement ...35

2.5.5 Monomer cleavage analysis ...35

Mass spectrometry...35

2.5.6 Crystal composition analysis...35

2.6APICALGROEL-Aβ ISOLATION AS AN OLIGOMER...36

2.6.1 Time dependent oligomerization analysis ...36

Superdex 75 size exclusion chromatography...36

Room temperature incubation ...36

Dynamic light scattering ...37

Western blotting ...37

2.7 THE HIGHER MOLECULAR WEIGHT PROTEIN...37

2.7.1 Non-specific protein or an ApicalGroEL-Aβ oligomer ...37

Western blotting ...37

2.8 GENERAL TECHNIQUES, MEASUREMENTS AND PREPARATIONS...37

2.8.1 DNA separation by agarose gel...37

2.8.2 Protein separation by SDS-PAGE ...38

2.8.3 Western blot analysis ...38

2.8.4 DNA concentration measurement...39

2.8.5 Protein concentration measurement ...39

2.8.6 Bacterial growth measurement...39

2.8.7 LB agar plates ...39

2.8.8 LB cultures...39

2.8.9 LB media...40

2.9 APPARATUS...40

CHAPTER 3; RESULTS ...41

3.1 GENERAL RESULTS...41

3.1.1 Cloning of the ApicalGroEL-Aβ constructs ...41

3.1.2 Expression of the ApicalGroEL-Aβ constructs...41

3.1.3 Purification of the ApicalGroEL-Aβ constructs ...42

3.2 APICALGROEL-Aβ ISOLATION AS A MONOMER...43

3.2.1 Crystallization ...43

3.2.2 Solved structures...44

A376-6SAG-Aβ42 structure 1...44

A376-6SAG-Aβ42 structure 2...45

A376-6MPT-Aβ42 structure ...46

A376-6SAG-Aβ17-42 structure ...47

3.2.3 Monomer cleavage analysis ...47

Mass spectrometry...47

Western blotting ...48

3.2.4 Crystal composition analysis...48

Western blotting ...48

3.3 APICALGROEL-Aβ ISOLATION AS AN OLIGOMER...49

3.3.1 Time dependent oligomerization analysis ...49

Dynamic light scattering ...49

Western blotting ...50

3.4 THE HIGHER MOLECULAR WEIGHT PROTEIN...51

3.4.1 Non-specific protein or an ApicalGroEL-Aβ oligomer...51

SDS-PAGE analysis ...51

Western blotting ...52

CHAPTER 4; DISCUSSION AND CONCLUSION...53

4.1 APICALGROEL-Aβ ISOLATION AS A MONOMER...53

4.2 APICALGROEL-Aβ ISOLATION AS AN OLIGOMER...54

4.3 THE HIGHER MOLECULAR WEIGHT PROTEIN...54

4.4 SUGGESTIONS FOR FURTHER RESEARCH...54

4.5 CONCLUDING STATEMENT...55

REFERENCES...56

WEBSITES...59

SUPPLEMENTS ...60

SUPPLEMENT 1 AMYLOID BETA PEPTIDE FRAGMENT STRUCTURES...60

SUPPLEMENT 2 GROEL BACKGROUND INFORMATION...62

SUPPLEMENT 3 CHROMATOGRAM OF A NICKEL COLUMN PURIFICATION...63

SUPPLEMENT 4 DYNAMIC LIGHT SCATTERING ANALYSIS...64

SUPPLEMENT 5 SDS-PAGE OF A SIZE EXCLUSION PURIFICATION...65

SUPPLEMENT 6 PRIMER NUCLEOTIDE SEQUENCES...66

SUPPLEMENT 7 LINKER NUCLEOTIDE SEQUENCES...68

SUPPLEMENT 8 AMINO ACIDS SEQUENCE CONSTRUCTS...70

Figure 1-2 Amyloid cascade³.

Figure 1-1 Senile plaques in Alzheimer’s disease. Seen with a silver stain⁴³.

Chapter 1; Introduction

1.1 Biochemical background of amyloid beta 1.1.1 Amyloid diseases

Protein misfolding diseases arise from the failure of a protein to adopt or remain in its native biological conformation. These incorrectly folded proteins often self-associate to form an aggregate, which can lead to the so-called aggregation diseases. There are approximately forty known aggregation diseases that result in amyloid formation, which is known as an amyloid disease or amyloidosis. The aggregation can occur in one single type of tissue or in multiple types of tissues. Amyloids are extracellular deposits of insoluble fibrils in organs and tissues. They bind the dye Congo red and give a green birefringence under a polarization microscope after Congo red staining. Congo red recognizes the β-pleated structure found in the fibrils, this is not well understood. Some amyloid diseases are: Parkinson’s disease, Huntington’s disease, Type II diabetes and the most well-known and well-studied Alzheimer’s disease (AD). In AD, aggregation and deposition of the protein amyloid β (Aβ) takes place in the human brain as extracellular senile plaques (Fig. 1-1)^1-3.

1.1.2 Alzheimer’s disease

Alzheimer’s disease is a neurodegenerative disease of the central nervous system. The first case was presented by the psychiatrist Alois Alzheimer in 1906. Now more than 30 million people worldwide are affected with AD⁴. Alzheimer’s disease patients experience gradual memory loss (dementia) and other declines in cognitive function. The disease is characterized by the presence of neuropathology hallmarks such as: senile plaques, intraneuronal neurofibrillary tangles (NFTs) and the loss of synapses and neurons in the brain.

Neurofibrillary tangles are composed of aggregated hyperphosphorylated protein tau. Whereas extracellular plaques are mainly composed of amyloid fibrils of the protein amyloid β (Aβ) (Fig. 1-1). Although amyloid β arises naturally in the human brain throughout life, it seems to play a central role in the pathogenesis of AD^4,5.

The neurotoxic species of Aβ was long thought to only be the insoluble fibrillar form found in extracellular plaques in AD brain. However, it was observed that the plaques are more prevalent with age in AD patients and they are also

found in cognitively normal individuals. In addition to that the number of plaques in the brain seemed to roughly correlate with the severity of the disease^6,7. Also, soluble Aβ in the brain is shown to be much better correlate with the severity of the disease. Soluble Aβ has the potential to do harm over a much wider area than insoluble Aβ due to insoluble Aβ being fixed to one single location while soluble Aβ is able to freely diffuse into the synaptic cleft^8,9.

This suggests a neurotoxic role for the soluble Aβ content in the brain. It is not known which assemblies of Aβ are pathogenic, although more and more is becoming clear. The disease is thought to develop through a process called amyloid cascade (Fig. 1-2) where an increase of Aβ level results in oligomerization and aggregation of Aβresidues 1-42 (Aβ42) and thereby triggering a downstream cascade of pathological events³. The deposition of the protein tau intraneuronal as NFTs is believed to be a downstream event in the amyloid cascade⁴.

1.1.3 Amyloid beta discovery and formation process

The link of Aβ with Alzheimer’s disease was first made by Glenner and Wong (1984).

They were the first to discover that the amyloid fibrils found in the brains of AD patients are mainly composed of a 4.5 kDa protein that they named amyloid fibril protein β¹⁰. The gene encoding for the precursor of amyloid β, amyloid precursor protein (APP) (an integral membrane protein) is located on chromosome 21¹¹.

Aβ arises naturally in the human brain throughout life as a result of metabolic processing of the amyloid precursor protein (APP). APP is expressed in many tissues, both neural and nonneural, with high expression in the brain¹². There are two competing pathways that can process APP. In the non-amyloidogenic pathway APP is cleaved by the protease α-secretase that cleaves APP in the middle of Aβ and thus prevents Aβ generation. In the amyloidogenic pathway APP is cleaved by the proteases β- and γ-secretase, which eventually results in liberation of the Aβ peptide into the extracellular space^13,14.

When APP is cleaved via the amyloidogenic pathway it is first cleaved by β-secretase (also known as β-site APP cleaving enzyme: BACE), resulting in secretion of the ectodomain (Fig. 1-3 A). Next an intramembrane cleavage takes place, which is mediated by the γ-secretase complex (Fig. 1-3 B). After the BACE cleavage the remaining membrane part of APP (CTFβ) is transferred to the active site of the γ-secretase complex, in between the transmembrane domains 6 and 7 of presenilin-1 (PS1) or PS2. PS1 and PS2 are autoproteases, after cleavage their N- and C-terminal fragments are created (NTF and CTF). In addition to PS1 and PS2, three other proteins make part of the γ-secretase complex; nicastrin (NCT), APH1 and PEN2. The γ-secretase complex cleaves CTFβ at multiple sites (Fig. 1-3 C). The γ-secretase complex ε-cleavage releases the APP intracellular domain (AICD) into the cytosol (Fig. 1-3 B, C and D). The remaining membrane fragment is next cut at the ζ-site and at last at the γ-site which releases Aβ into the extracellular space (Fig. 1-3 B, C and D)³. The γ-cleavage site is variable; it happens after amino acid 38, 40 or 42, producing different lengths of full-length Aβ peptides (39- 43 amino acids) (Fig 1-3 C and D), which has a large influence on the pathogenicity of Aβ³.

Figure 1-3 Amyloid β generation by metabolic processing of the amyloid precursor protein³.

The 40 residue peptide, Aβ40 is the major Aβ peptide found in the cerebrospinal fluid (90

%), while the 42 residue peptide, Aβ42 is the major component of the plaques in AD (see Fig. 1-4 for the amino sequence of full-length Aβ peptides)¹⁵. Aβ42 is much more pathogenic than Aβ40. It is more hydrophobic compared to Aβ40 because of its two extra C-terminal residues; alanine and isoleucine (Fig. 1-4). It therefore aggregates more rapidly and forms stable oligomers faster and easier than Aβ40. Because of these properties,Aβ42 is thought to be more pathogenic and amyloidogenic than Aβ403.

Imbalance between the Aβ production and clearance resulting in oligomerization and aggregation of Aβ42 into plaques is thought to be an important event in AD. More than a hundred mutations in the gene encoding for the precursor of Aβ, APP and mutations in genes encoding for PS1 and PS2 have been found to cause AD. These mutations increase the production of Aβ and especially the more amyloidogenic Aβ423.

1.1.4 Amyloid beta fibrillization process

There are multiple Aβ assemblies that can be divided into two distinct categories: fibrillar and non-fibrillar. The non-fibrillar Aβ assemblies include the Aβ monomer and small soluble oligomers; soluble oligomers are assemblies of Aβ monomers. The fibrillar Aβ assemblies include protofibrils and fibrils.

NH2 – D A E F R H D S G Y E V H H Q K L V F F A E D V G S N K G A I I G L M V G G V V I A – COOH 10 20 30 40 42

Figure 1-4 Amino acid sequence of the full-length Aβ peptides: Aβ40 and Aβ42. Aβ42 includes two extra C-terminal residues (shown in gray).

Figure 1-5 Aβ42 assembly model. Monomers assemble into paranuclei that assemble further to form large oligomers. Monomers, paranuclei and large oligomers are predominately unstructured. Protofibril may form from large oligomers. There are two alternative pathways in which direct addition of monomers or paranuclei to protofibrils is suggested (dotted arrows). Protofibrils maturate into insoluble fibrils. α = helical elements. β = β-sheet/β-turn U = unstructured¹⁹.

The Aβ monomer has an α-helical structure in membrane-like conditions and adopts an unordered conformation (collapsed coil) in aqueous solution (section 1.1.8). In contrast Aβ in fibrils has a β-sheet structure (section 1.1.8). Therefore a conformational change to the β-sheet structure is thought to result into fibril formation¹⁶.

Monomers assemble rapidly to form oligomers and equilibrium is formed¹⁷. Once the oligomers are starting to interact slowly to form higher order assemblies such as paranuclei (spherical oligomers) and large oligomers, protofibrils are formed; small elongated Aβ oligomers with β-sheet structure (Fig. 1-5)¹⁸. The protofibril formation is reversible on dilution; a protofibril can depolymerize into oligomers and monomers. Thus equilibrium is formed between protofibrils and oligomers.

Protofibrils maturate into insoluble fibrils with β-sheet structure¹⁹. The formation of fibrils is irreversible; once it is formed it is not easily degraded into lower order species even when diluting the fibrils⁷.

It is not completely clear how fibril formation occurs. There is evidence that oligomer formation is not even necessary for fibrillization, suggesting that oligomer and fibril formation are two distinct pathways¹⁷. Fibrils can be composed of Aβ40 as well as Aβ42, but are mainly composed of Aβ42. The much higher propensity of Aβ42 to aggregate into insoluble fibrils compared to Aβ40 is due to the two extra C-terminal amino acids in Aβ42. These two amino acids appear to be important for fibril formation kinetics²⁰.

1.1.5 Amyloid beta oligomer pathology

Intraneuronal accumulation of Aβ precedes the appearance of senile plaques and NFT’s (Fig. 1-2). The intracellular Aβ is neurotoxic and affects synaptic function. The intracellular Aβ consists of monomers and oligomers²¹. A large variety of Aβ oligomers sizes have been found in in vivo and in vitro studies (Tab. 1-1). Oligomer size distributions of Aβ40 and Aβ42 are shown to be distinct, large oligomers are only observed for Aβ4219. It is unclear what size Aβ oligomers exist in vivo and which ones are pathological. However, there are studies pointing towards toxicity of specific oligomeric states.

The neurotoxicity of soluble Aβ42 was determined in an in vitro study where synthetic Aβ42 (forming monomer as well as oligomer assemblies) was injected into the cytoplasm

Oligomer assembly Origin Dimer: Aβ40 andAβ42 Synthetic¹⁹

In vivo human²⁵ In vitro4,23, 25,26

Trimer: Aβ40 andAβ42 Synthetic¹⁹ In vitro 4,23,25,26

In vivo transgenic mice²⁴ Tetramer: Aβ40 and Aβ42 Synthetic^4,19

Pentamer: Aβ42 Synthetic¹⁹ Hexamer: Aβ42 Synthetic¹⁹

In vivo transgenic mice²⁴ Octamer: Aβ42 Synthetic¹⁹

Nonamer: Aβ42 Synthetic¹⁹

In vivo transgenic mice²⁴ Dodecamer: Aβ42 Synthetic^19,27

In vivo transgenic mice²⁴ In vivo human²⁷

Octadecamer: Aβ42 Synthetic¹⁹ Paranuclei: Aβ40 and Aβ42 Synthetic^19,28

Table 1-1 Oligomer assemblies of Aβ and their origin.

of cultured primary human neurons and caused significant cell death. Aβ40 was significantly less toxic and the toxicity of Aβ42 was proven to be selective for neurons²². One of the studied neuropathological events caused by human Aβ oligomers is the blocking of hippocampal long-term potentiation (LTP) in rats. LTP is blocked in rats in vivo when the oligomers are injected into the hippocampus. LTP is considered to be an important mechanism for learning and memory. It is possible that the memory decline seen in AD is a result of similar LTP impairment as is caused by Aβ oligomers seen in rats²³.

In vitro experiments on mice hippocampal brain slices showed the strongest LTP inhibition caused by Aβ trimers and a lower inhibition caused by dimers and tetramers⁴. In contrast to the oligomer assemblies, both the above in vivo and in vitro studies show that monomeric Aβ does not affect LTP. Suggesting a pathological role for Aβ oligomer assemblies.

A specific oligomeric state that was found to be toxic in an in vivo study is a dodecamer.

In transgenic mice memory impairment is shown to be correlated with accumulation of an extracellular 56 kDa oligomer (Aβ*56) (dodecamer). Purified Aβ*56 from mice brains disrupts memory when injected into the brains of rats²⁴.

1.1.6 Amyloid beta oligomer kinetics

The smallest oligomer sizes of Aβ, dimers and trimers, are very stable; they are not altered by SDS or 8 M urea^23,24. In contrast, the oligomerization process itself is very sensitive to denaturation conditions.

Chen and Glabe (2006) studied the oligomer kinetics of Aβ40 and Aβ42 in vitro. This study shows that Aβ42 oligomerization is time and concentration dependent. They incubated the two peptides at 50 µM at room temperature in different concentrations of urea (Fig. 1-6). In 0.2 M urea oligomers begin to appear around 16-52 hours of incubation and increased with longer incubations (Fig 1-6). Aβ42 oligomerization is

Figure 1-6 Oligomer formation kinetics.

Samples at different time points were dotted on the nitrocellulose membrane. The membrane was probed by oligomer-specific antibody A11 without boiling. The spotted time is indicated above the blot and the urea concentration is indicated on the left side of the blot¹⁷.

concentration dependent, trimers and tetramers are formed at 25 µM and 50 µM but not at 12.5 µM. Aβ40 did not assemble as trimers and tetramers. Urea concentrations >0.2 M already inhibited oligomer formation. Thus the appearance of oligomers requires native folded proteins to start with (a high enough concentration of urea can disrupt the native conformation of a protein)¹⁷.

1.1.7 Common fibril and oligomer structures of amyloids

The different amyloidogenic proteins found in amyloid deposits do not share apparent primary structure similarity. Although the oligomers and fibrils of the proteins are unrelated in primary sequence they do share similarity in their secondary structure, suggesting that the pathogenicity of amyloidogenic proteins is structure related and not sequence related².

The fibrils are long and unbranched and are composed of finer fibrils: the protofilaments.

Amyloid fibrils can be ~10 nm in diameter and a micrometer in length, while the protofibrils are up to 150 nm in length and 5 nm in diameter (Fig. 1-5)²⁹. As stated above, amyloid fibrils have a common structural feature: the cross-β unit structure. X-ray fiber diffraction for different amyloid fibrils shows a cross-β diffraction pattern. There is a meridional reflection at ~4.7 Å and an equatorial reflection at ~ 10 Å (Fig. 1-7 A). This diffraction pattern is produced by the organization of specific parts of the peptides in the fibril as the cross-β structure³⁰.

The cross-β structure (Fig. 1-7 B) is composed of parallel, antiparallel or parallel and antiparallel β-sheets along the fibril axis. The intersheet spacing ranges from 5-14 Å. The β-strand segments in the β-sheets run perpendicular to the fibril axis by hydrogen- bonding up and down the sheet to identical molecules. The interstrand spacing is 4.7 Å.

The structure and organization of the proteins in the fibrils is different for each amyloidogenic protein and have to be defined individually. Unfortunately fibrils have a noncrystalline insoluble nature thereby making structural studies very difficult³⁰.

A common structural feature of the oligomer assemblies of amyloidogenic proteins is also suggested, since a study done by Kayed et al. (2003) was able to produce an oligomer-specific antibody that could recognize different amyloidogenic proteins. This suggests the presence of a shared structure in the different amyloidogenic proteins. In addition, this antibody can neutralize the pathogenicity of the oligomers in vitro, indicating a common pathological mechanism for all amyloidogenic proteins associated with this common conformation³¹.

1.1.8 Amyloid beta proposed structures

Conformation transition from α-helical to β-sheet

Aβ aggregation is thought to be the result of a conformational transition from a mainly α-helical to a β-sheet conformation: the helix-coil-sheet progression¹⁵. The Aβ peptide in vivo is helical when it is part of APP and it has a β conformation in plaques¹⁶.

The conformation transition is hypothesized to be as followed: in the cell membrane when Aβ is part of APP, it has an α-helical conformation³². After cleavage and release in the extracellular space, the Aβ monomer forms a collapsed coil²² and starts to self- assemble into oligomers with still unknown conformations and sizes. The oligomers assemble further into protofibrils with β-sheet conformation and the protofibrils assemble into β- sheet fibrils¹⁹.

Figure 1-7 Cross-β structure of amyloid fibrils. (A) Drawing of the cross-β diffraction. (B) Cross-β structure of amyloid fibrils³⁰.

Figure 1-8 NMR structures of Aβ in different environments. (A) α-Helical structure of Aβ42 in apolar environment³². Image generated in Jmol from 1IYT. (B) Structure of Aβ10-35 in water as a collapsed coil²². Image generated in Jmol from 1HZ3.

Amyloid beta monomer α-helical structures in apolar solution

There are many different studies that have studied solution structures (of different lengths) of Aβ peptides in apolar environment, such as mixtures of organic solvents with water and micellar solutions. The studies generally do not agree about the structure of Aβ in apolar solution, suggesting that the secondary structure of Aβ is strongly dependent on experimental conditions³². However the studies do agree that Aβ is mainly in an α-helical conformation in apolar environment (Fig. 1-8 A). This is consistent with the hypothesis that Aβ has an α-helical conformation in apolar environment in vivo; namely within the cell membrane¹⁶.

Amyloid beta monomer collapsed coil structure in polar solution

In water, Aβ adopts a collapsed coil that has no α-helical or β-sheet structure (Fig. 1-8 B). Aβ forms this conformation because it is more compact and has better solvation thermodynamics than the α-helical conformation²².

Amyloid beta fibril β-sheet structures

Aβ fibrils form β-sheet structure³³. A few fibril structure models of full-length Aβ peptides have been proposed so far.

A fibril model for Aβ1-40 based on solid state nuclear magnetic resonance (ssNMR) is proposed by Petkova et al. (2002). The ssNMR data suggests that residues 1-8 are disordered and they were omitted. The proposed structure consists of two β-strands (residues 9-24 and 30-40) separated with a bend formed by residues 25-29 (Fig. 1-9 A).

The β-strands interact through side chain interactions. The side chains of residues D23 and K28 form a salt bridge. The β-strands form two in-register parallel β-sheets: the cross-β unit. The β-sheets run perpendicular to the fibril axis (Fig. 1-9 B). Two of these layers form a protofilament: the narrowest Aβ fibril. Two or more protofilaments pack together to form thicker fibrils (Fig. 1-9 C)³⁴.

Figure 1-9 ssNMR model for Aβ1-40. (A) Aβ1-40 molecule, showing the two β-strands and bend³⁰. (B) The cross-β structure made of two in-register parallel β-sheets³⁴. (C) Lateral association of protofilaments to form thicker fibrils³⁰.

A comparable fibril model for Aβ1-42 has been proposed by Lührs et al. (2005). In this study they used quenched hydrogen-exchange NMR (qHX-NMR) on recombinant

35MoxAβ1-42. They suggest that residues 1-16 are disordered and that the two β-strands consist of residues 18-26 and 31-42 and the loop region of residues 27-30. The side chains of residues D23 and K29 form a salt bridge³⁵.

A slightly different model for Aβ40 using proline scanning mutagenesis has been proposed by Williams et al. (2004). Residues 1-14 and 37-40 are suggested to be disordered. Residues 15-21, 24-28 and 31-36 form β-strands (3 β-strands instead of two) and residues 22-23 and 29-30 form turns (Fig. 1-10)²⁰.

1.1.9 Drugs against Alzheimer’s disease

There are multiple possible targets to selectively inhibit the production, oligomerization or fibrillization of Aβ. It is important to determine which species are responsible for pathogenesis, because inhibiting fibrillization at a too late stage can cause accumulation of the pathogenic species and thereby accelerates the disease instead of preventing it⁷. One could inhibit the liberation of Aβ from its precursor by inhibition of the secretases.

This can be a dangerous approach with unwanted side effects. The secretases have enzymatic activities and process many important substrates. Thus a drug has to allow some activity of the secretases^3,12. Second, one could try to divert APP to follow the nonamyloidogenic processing pathway instead of the amyloidogenic pathway¹². Last, one could attempt to prevent oligomerization of Aβ with Aβ immunotherapy. Clinical trials with AD patients resulted in some promising results. Unfortunately some patients developed unwanted inflammatory reactions. One can also attempt to prevent oligomerization of Aβ with drugs that target the monomer and prevent oligomerization.

The oligomers can also be targeted to prevent their pathogenicy³⁶.

There are a number of approaches that could be useful in AD therapies. Further characterization of the amyloid beta cascade is necessary to design safe and effective therapeutic drug therapies¹⁴.

Figure 1-10 Proline scanning mutagenesis protofilament model for Aβ1-40. The 15-36 residues of the Aβ molecules are stacked in the direction parallel to the fibril axis²⁰.

1.2 Protein crystallography

The goal of protein crystallography is the production of a high-resolution molecular model to gain information about a protein its three-dimensional structure. The knowledge of a protein structure reveals insight into the protein its function. Interactions with other molecules can be studied, for example better understanding about how enzymes catalyze metabolic reactions, how they switch from an inactive to an active state by changing their conformation or how a protein binds to DNA, can be gained. With this information drugs targeting a specific part of the protein can be designed to inhibit unwanted protein functions. To obtain a molecular model using protein crystallography, the diffraction pattern of scattered X-rays caused by many highly ordered identical molecules in a protein crystal must be obtained and interpret. X-rays are used because their wavelength is small enough to be diffracted by the electron clouds surrounding a molecule. A crystal containing many ordered identically orientated protein molecules (which diffract identically) is necessary to produce strong enough diffracted X-ray beams that can be detected; the diffraction from a single molecule is too weak to be detected³⁷.

1.2.1 Crystallization

To be able to solve the structure of a protein, the protein must be expressed in large quantity and purified; a high-quality crystal must be obtained under optimal crystallization conditions. The protein solidifies into a crystalline state by the arrangement of the molecules in an ordered three-dimensional array (Fig. 1-11). A crystal is an array of many unit cells packed together to form a crystal; a unit cell is the smallest repeating element in a crystal. So knowing the content of one unit cell is like knowing the content of the whole crystal³⁷. The unit cell is described by the so-called lattice parameters:

the length of the cell edges (a, b and c) and the angles between them (α,β and γ). The position of the atoms inside the unit cell is described by atomic positions measured from a lattice point (xx,yyzz). These parameters make it possible to group crystal structures in crystal systems. For example crystals belonging to the orthorhombic crystal system have the lattice parameters: a ≠ ^b ≠ c, and α, β and γ are always 90°⁴⁴.

Typical crystallization conditions contain buffer, precipitant and salt. Under favorable conditions a crystal will form, but most often precipitate or salt crystals will form or nothing happens at all. Successful crystallization is dependent on the salt, protein and precipitant concentrations, and pH and temperature. Finding the right condition takes a considerable trail and error and often many conditions must be tried before a condition that produces crystals is found, if any at all. Once a condition containing a good crystal is found, optimization of this condition (making small variations) is needed to produce high-quality crystals³⁷.

Figure 1-11 Six unit cells, each unit cell here consists of two alanine molecules. a, b and c are the edges of the unit cell³⁷.

Crystals can form in a few days, several weeks or even months, it is therefore important to continuously check the crystal plates for crystal formation. There are different crystallization methods including hanging-drop (in which the protein/crystallization solution drop hangs above a reservoir containing the crystallization solution) and sitting- drop methods (Fig. 1-12). There are crystallization robots available that can quickly set up systematically varied conditions. This is often used to get a quick impression of the crystallization

conditions that promotes crystallization³⁷. In the mainly used vapor diffusion technique, equilibrium between a protein/crystallization solution drop and a larger reservoir containing only crystallization solution is reached. Crystals are grown by slow precipitation from an aqueous solution. Water starts to evaporate slowly from the protein/crystallization solution drop to the reservoir until the precipitant concentration is the same in both solutions. The evaporation increases the protein and precipitant concentration, which promotes the first stage of crystal formation: nucleation. Nucleation is the formation of groups of molecules from which crystals start growing³⁷.

1.2.2 X-ray diffraction

Protein crystals are very fragile, that is because the molecules in the crystal are mainly held together by hydrogen bonds; non-covalent interactions. There are many different forms and sizes of crystals, all with different diffraction quality. Preferably only nicely formed crystals with sharp edges and smooth surfaces are screened on the X-ray machine for diffraction quality³⁷.

To prevent damaging the crystal by the X-ray beam, the crystal is kept flash frozen in a liquid nitrogen stream. Freezing a crystal can also result in damage due to the formation of ice crystals, therefore the crystal is dipped in a cryoprotectant an ice-preventing agent. The crystal is removed from its mother liquor (the original crystallization solution) by picking it up in a circular loop of glass wool or synthetic fiber. Next the crystal is dipped in cryoprotectant and placed onto the goniometer of the X-ray source where it is kept in a constant stream of liquid nitrogen (120 K) to

Figure 1-12 The sitting-drop method showing a well- plate, in which 24 sitting-drop crystallization trails can be carried out. Each well contains a pedestal with a concave top, in which the protein/crystallization drop sits. Vapor diffusion occurs between the drop and the reservoir containing buffer, precipitant and salt³⁷.

Figure 1-13 X-ray diffraction. The electron clouds in the crystal scatter the X-ray beam, producing diffracted X- rays each of which produces a spot (reflection) that can be detected³⁷.

flash-freeze it³⁷. When the X-ray beam strikes the crystal, the X-ray will be scattered by the electron clouds surrounding the molecules in the crystal and produce a diffraction pattern of spots known as reflections detected by an X-ray detector (Fig. 1-13). The diffraction pattern relates to atom positions in the molecules, while the intensity of each spot in the diffraction pattern correlates how strongly each atom diffracts in the molecule;

this information is needed for solving protein structure along with known phase information. Depending on the crystal quality sharp spots should be visible and the crystal should diffract to at least 3 Å (Ångstrom) to be able to obtain an interpretable electon-density map. If that is the case, collecting a full data set can be considered³⁷. 1.2.3 The phase problem

To be able to calculate the electron density based on the diffraction pattern three parameters of each reflection must be measured: the amplitude, frequency and phase. The amplitude and frequency are accessible in the data that is obtained, while the phase is not.

The phase angle for each reflection has to be determined. It can be determined with three different techniques: isomorphous replacement, anomalous scattering or molecular replacement³⁷.

In the isomorphous replacement approach, a heavy atom is added, thereby changing the diffraction pattern compared to the diffraction pattern of the native crystal. The change in diffraction pattern can be used to obtain estimates of the phase angle³⁷.

Anomalous scattering is also based on adding a heavy atom. Heavy atoms have an absorption edge near the wavelength of X-rays³⁷. Collection of three data sets from the same crystal at different wavelength around the absorption edge of the anomalous scatterer makes it possible to determine the phase⁴⁵.

Molecular replacement makes use of a phasing model to determine the structure of the new protein. The phasing model is a known homologous protein with at least 25%

sequence identity. The phases can be calculated by placing the model of the known protein in the unit cell of the new protein³⁷.

1.2.4 Model building

Once an electron-density map is obtained a molecular model can be built into the density.

The model must be in agreement with the principles of molecular structure and stereochemistry and must fit into the electron density³⁷. To improve the model it can be refined against the data to improve the phases, which results in a clearer map and a clearer model. This is done to make the model in better agreement with the data. This cycle is repeated several times until no further improvement is made and hopefully an accurate model results^37,45. Occasionally portions of the known sequence of a protein cannot be found back in the electron-density map. That can be because the region is disordered or flexible. It is also not uncommon for termini residues to be missing from the model³⁷.

1.3 Aim and procedure

More and more evidence indicates that especially the soluble Aβ oligomers are neurotoxic and play an important role in the Alzheimer’s disease related pathology^4,5,23,25. Therefore the ultimate aim of this study is to solve the structure of human Aβ42 in an oligomeric state for rational structure based drug design against Alzheimer’s disease.

Protein purification and X-ray crystallography techniques are being used. Due to the extreme insolubility of Aβ it is hard to separate and thereby purify it from the other proteins in solution. All the crystallography structures proposed so far are fragment peptides of Aβ³⁸ (Sup. 1).

To be able to study Aβ with X-ray crystallography, Aβ was previously linked to the full- length chaperonin protein GroEL at the last residue of GroEL residue D523 (Fig. 1-14) (see Sup. 2 for background information of GroEL). It is difficult to trap Aβ in an oligomeric state because of its hydrophobic nature and high propensity for aggregation.

Protecting Aβ inside the tunnel of GroEL will limit its oligomerization and thus its aggregation. GroEL forms a heptamer with an internal cavity in which Aβ was situated.

In this manner, Aβ was prevented from self-association into insoluble fibrils; it can only form oligomeric states up to heptamers. Unfortunately, this specific protein complex assembly has not resulted in the structure of Aβ yet.

In this study Aβ is linked to the apical domain of GroEL (residues 191-376 of GroEL) (Fig. 1-15), creating the fusion protein ApicalGroEL-Aβ. In the fusion protein, Aβ is fused to the C-terminal end of the apical domain of GroEL (residue V376). This approach should allow for more flexibility, as Aβ is no longer trapped in a tunnel. At the same time, it inhibits the fibrillization and prevents Aβ precipitation, thereby forming the

Figure 1-15 Apical domain of GroEL. In blue apical residues E191-V336. In red the last two helices of the apical domain, residues G337- V376. Image generated in Chimera from PDB ID code: 1KID.

Figure 1-14 Center view of the full-length GroEL monomer. Aβ42 was linked to residue D523.

Image generated in SPOCK from the structure with Protein Data Bank identification (PDB ID) code: 1GRL.

possibility to trap Aβ as an oligomer. The apical domain of GroEL does not have the propensity to oligomerize; therefore any oligomer formation is thought to be Aβ dependent.

Multiple ApicalGroEL-Aβ fusion protein constructs are built, with a general construction (Fig 1-16). They contain a N-terminal His-tag followed by the apical domain of GroEL.

In between the apical domain and Aβ a synthetic sequence, the so-called linker is placed.

The constructs vary in the following three characteristics.

Different linkers are placed in between GroEL and Aβ to manipulate the arrangement of Aβ. The tested linkers are distinguished from each other by their length and hydrophobicity. Also, different size combinations of the apical domain of GroEL and Aβ are tested. Regarding the apical domain of GroEL constructs with and without residues 337-376 are tested; A191-376 and A191-336. A191-336 does not contain the last two helices that are present in A191-376 (Fig. 1-15). This is done to manipulate the packing of GroEL and Aβ in the unit cell. It is known that the first 16 residues of Aβ are disordered¹⁶. Therefore, in this study constructs with and without residues 1-16 are tested; Aβ1-42 and Aβ17-42.

In summary, different combinations of apical GroEL, linkers and Aβ are built, the ApicalGroEL-Aβ constructs are expressed and purified and the monomer and different oligomeric states present are also separated for crystallization purposes. All together these constructs are made with the hopeful expectation that one or more of them result in the formation a conformation of Aβ linked to GroEL as an oligomer that produces well diffracting crystals.

Two different crystallization approaches are used. The first approach is to isolate the protein as a monomer and screen for crystallization. Since Aβ has the propensity to form oligomers when it is at a high enough concentration¹⁷, the assumption is made that the protein will oligomerize as a result of concentration and crystallize as an oligomer. The second approach is to isolate ApicalGroEL-Aβ in an oligomeric state and to crystallize the protein in this oligomeric state. The protein monomer is purified, brought to high concentration and incubated to see if it will form oligomers in a time dependent manner, which can be purified and crystallized.

On our way to well diffracting crystals the following questions are tried to be answered:

• Do the constructs that vary in the size of the apical domain, linker length and hydrophobicity, and Aβ truncation show specific properties compared to each other?

• In what state(s) (monomeric and/or oligomeric) is it possible to purify and crystallize the protein?

• Is the monomer that is observed composed of the whole ApicalGroEL-Aβ protein or is it being cleaved?

• Is Aβ still present in the crystals or is Aβ cleaved off and are the crystals made of cleaved ApicalGroEL-Aβ protein?

Figure 1-16 The fusion protein ApicalGroEL-Aβ.

• Is it possible to assemble the protein to an oligomeric state in a time-dependent manner?

• What are the approximate sizes of the oligomers that are observed?

• Are the observed oligomers composed of the whole protein or are the oligomers formed of cleaved protein?

• Which ApicalGroEL-Aβ protein construct(s) is/are best to use for our aim?

• Are there ApicalGroEL-Aβ protein constructs as a monomer or as an oligomeric state that produce well diffracting crystals?

Different techniques are used to answer these questions. Purification techniques such as affinity and size exclusion chromatography are used to purify the different protein assemblies. Sitting drop methods are used to crystallize ApicalGroEL-Aβ. For molecular weight analysis, size exclusion chromatography and mass spectrometry are used.

Dynamic light scattering is used to monitor oligomer formation and western blots are used to analyze the composition of the protein.

Eventually the most crucial question has to be answered:

Is it possible by solving the structure of Aβ in an oligomeric state to gain better understanding of the Aβ mechanism in order to design drugs against Alzheimer’s disease?

Figure 2-1 ApicalGroEL-Aβ gene construction, also showing four specific restriction sites.

Chapter 2; Materials and methods

2.1 General experiments: plasmid, gene and protein constructs

The ApicalGroEL-Aβ constructs that were built had a general construction (Fig. 1-16 and 2-1). The constructs had a N-terminal polyhistidine-tag (6X-His-tag) that has affinity for nickel ions that can be used for affinity purification. Followed by the apical domain of GroEL, the linker region and most C-terminal Aβ. The amino acid sequences and molecular weights of the built constructs can be found in supplement 8. The vector that was used to insert and express the genes of interest in is pET-28b [Novagen]. The linkers were ordered directly from Integrated DNA Technologies. The linker nucleotide sequences can be found in supplement 7. The sequence of the apical domain of GroEL was in a previous study amplified from the E.coli genome (strain K12). Aβ1-42 was in a previous study amplified from human brain cDNA. The ApicalGroEL-Aβ constructs listed in table 2-1 were built using the molecular cloning techniques as described in the next section.

Construct name GroEL residues Linker (amino acid sequence)

Aβ residues

A376-SAG-Aβ42 Apical 191-376 SAG Aβ 1-42

A376-6SAG-Aβ42 Apical 191-376 SAGSAG Aβ 1-42

A376-9SAG-Aβ42 Apical 191-376 SAGSAGSAG Aβ 1-42

A376-12SAG-Aβ42 Apical 191-376 SAGSAGSAGSAG Aβ 1-42

A376-6MPT-Aβ42 Apical 191-376 MPTATA Aβ 1-42

A376-9NSQ-Aβ42 Apical 191-376 NSQPNTNGS Aβ 1-42

A376-12NSS-Aβ42 Apical 191-376 NSSGSGSNSSGS Aβ 1-42

A376-6SAG-Aβ17-42 Apical 191-376 SAGSAG Aβ 17-42

A336-6SAG-Aβ42 Apical 191-336 SAGSAG Aβ 1-42

A336-12GSA-Aβ42 Apical 191-336 GSAGSAAGSGEF Aβ 1-42

A336-6SAG-Aβ17-42 Apical 191-336 SAGSAG Aβ 17-42

A376 control Apical 191-376 A336 control Apical 191-336 Table 2-1 ApicalGroEL-Aβ constructs.

2.2 General experiments: cloning of the ApicalGroEL-Aβ constructs 2.2.1 Plasmid preparation

Double digestion

To digest 1 µg (see section 2.8.4 for DNA concentration measurement) of plasmid pET- 28b, 5 µl (10x) Buffer 2 [New England Biolabs] and 20 Units (U) of both of the appropriate restriction enzymes [New England Biolabs] were added to 1 µg plasmid.

MilliQ water (mQ) was added to get an end volume of 50 µl. The chosen restriction enzymes were depending on which gene/linker had to be inserted (Fig. 2-1 and Tab. 2-2).

The reaction mixture was incubated at 37 °C for 3 hours. The DNA was separated by size with an agarose gel (see section 2.8.1 for DNA separation by agarose gel).

DNA extraction from agarose gel

The DNA was visualized under UV light and the double digested plasmid DNA band was cut out and the weight of the gel slice was measured. The DNA was extracted using the QIAquick Gel Extraction Kit [Qiagen] as followed.

Three volumes of Buffer QG were added to one gel volume (100 mg for 100 µl) and this was incubated in a heating block at 50 °C for 10 min until the gel slice was completely dissolved. To help dissolving the gel, the tube was vortexed every few minutes. Next one gel volume of isopropanol was added and mixed carefully. The solution then was applied to a Qiaquick spin column with collection tube and centrifuged for one minute at 13,200 rpm. The flow-through was discarded and applying 750 µl of buffer PE to the column and centrifuging for another minute washed the column. After discarding the flow- through the column was again centrifuged to make sure that all residual ethanol from Buffer PE was removed. Next the spin column was placed in an eppendorf tube and for elution of the DNA 30 µl mQ was directly applied to the center of the column membrane, incubated for five minutes and centrifuged for one minute. To remove all traces of ethanol the purified DNA was speedvaced for 10 minutes. The sample was now ready for ligation and could be used immediately or stored at -80 ºC for later use.

Gene or linker to insert in pET28b

Restriction enzyme 1

Restriction enzyme 2

Forward primer Reverse primer Apical 191-376 NdeI NheI NdeI apical Apical 376 NheI Apical 191-336 NdeI NheI NdeI apical Apical 336 NheI Apical 191-376 control NdeI XhoI NdeI apical Apical 376 XhoI Apical 191-336 control NdeI XhoI NdeI apical Apical 336 XhoI

SAG NheI HindIII NheI SAG Aβ XhoI

SAGSAG NheI HindIII NheI SAGSAG Aβ XhoI

SAGSAGSAG NheI HindIII NheI SAGSAGSAG Aβ XhoI

SAGSAGSAGSAG NheI HindIII NheI SAGSAGSAGSAG Aβ XhoI

MPTATA NheI HindIII NheI MPTATA Aβ XhoI

NSQPNTNGS NheI HindIII NheI NSQPNTNGS Aβ XhoI

NSSGSGSNSSGS NheI HindIII NheI NSSGSGSNSSGS Aβ XhoI GSAGSAAGSGEF NheI HindIII NheI GSAGSAAGSGEF Aβ XhoI

Aβ 1-42 HindIII XhoI HindIII Aβ1-42 Aβ XhoI

Aβ 17-42 HindIII XhoI HindIII Aβ17-42 Aβ XhoI

Table 2-2 Genes and linkers respective restriction enzymes and primers.

2.2.2 Insert preparation Linker preparation

The designed oligonucleotides were ordered from Integrated DNA Technologies, one of them included a 5’ partial NheI restriction site while the other one had a 5’ partial HindIII restriction site (Fig. 2-2). The forward and reverse oligonucleotides were both diluted to 100 µM with mQ to a final volume of 100 µl. These solutions were mixed together and vortexed to get a 200 µl duplex solution (100 µM). This was split into two PCR tubes and the following theromocycler annealing protocol was started:

After finishing the reaction the linker duplex was ready for ligation and could be used immediately or stored at -80 ºC for later use.

Gene preparation

The inserts were made by amplifying from plasmid constructs containing the gene of apical domain of GroEL and the gene of Aβ42. These inserts were: apical 191-376, apical 191-336, Aβ 1-42 and Aβ 17-42. See supplement 6 for primer nucleotide sequences and additional information.

PCR amplification

The designed primers were ordered from Integrated DNA Technologies. The primers were brought into solution by adding 1 µl mQ per nM primer making it a 1mM primer solution. This was diluted further to a 10 µM primer solution by adding 2 µl of 1 mM primer solution to 198 µl mQ.

A 300 µl PCR reaction mixture was prepared containing 6 µl (10 µM) of both of the appropriate forward and reverse primers (Tab. 2-2), 6 µl plasmid construct containing the desired gene and 282 µl PCR SuperMix High Fidelity [Invitrogen]. This was split into 50 µl reactions in PCR tubes and the following PCR protocol was started:

After finishing the reaction a few microliters were taken and ran on an agarose gel to confirm specific amplification. After confirmation the sample was ready for a PCR product purification to remove the primers and enzyme and could be used immediately or stored at –80 ºC for later use.

Figure 2-2 Linker construct. Two oligonucleotides annealed to each other. The red overhang contains a partial NheI restriction sequence. The blue overhang contains a partial HindIII restriction sequence.

94 ºC Tm + 10 ºC Tm + 5 ºC Tm – 2 ºC Tm – 7 ºC Tm – 12 ºC Tm – 20 ºC 4 ºC 2 min 5 min 5 min 2 min 2 min 2 min 10 min ∞

PCR product purification

To remove the primers and enzyme three volumes of Buffer PB [Qiagen] were added to one volume of PCR reaction mixture, this was carefully mixed and applied to a Qiaquick spin column [Qiagen] with collection tube. This was centrifuged for one minute at 13,200 rpm and the flow-through was discarded. Applying 700 µl of Buffer PE [Qiagen], incubating it for a few minutes and centrifuging for another minute washed the column.

After discarding the flow-through the column was centrifuged for 5 minutes to make sure that all residual ethanol from Buffer PE was removed. Next the QIAprep spin column was placed in an eppendorf tube and for eluting the DNA 100 µl mQ was directly applied to the center of the column membrane incubated for five minutes and centrifuged for one minute. Next another 50 µl of mQ was applied to the membrane, incubated and centrifuged. To remove all traces of ethanol the purified DNA was speedvaced for 10 minutes. The sample was now ready for further use and could be used immediately or stored at -80 ºC for later use.

Double digestion

The digestion protocol for the inserts is the same as is used for plasmid digestions, digesting 1 µg in a 50 µl reaction volume (see section 2.2.1).

Digested PCR product purification

The purification after digestion is the same as the PCR product purification protocol except for the elution step. The DNA was eluted first with 30 µl mQ and next with another 20 µl more mQ. The sample was now ready for ligation and could be used immediately or stored at -80 ºC for later use.

2.2.3 Ligation and transformation Linker ligation into cut plasmid

10 µl of linker (100 µM) and 5 µl of the double digested plasmid with NheI and HindIII were added together and mixed slightly. Of the DNA Dilution Buffer [Roche Applied Science] 2.5 µl was added to the tube. Next 10 µl of DNA Ligation Buffer [Roche Applied Science] was also added to the tube. Finally 2 µl of T4 DNA Ligase [Roche Applied Science] was added, the reaction was mixed carefully and the reaction was incubated at room temperature for 10 minutes. The ligation reaction was now ready for the transformation.

Gene ligation into cut plasmid

A slightly different protocol was used to ligate the bigger inserts apical 191-376, apical 191-336, Aβ 1-42 or Aβ 17-42 into the double digested plasmid (digested with the correct enzymes depending on which insert had to be inserted (Tab 2-2)).

A molar ratio of vector DNA to insert DNA of 1:5 was used. Of the plasmid 100 ng was used for the reaction. Depending on the molar ratio and size of the plasmid and the insert, the required amount of insert DNA was calculated (in ng). After adding the right amount of plasmid and insert DNA together, 2 µl of DNA Dilution Buffer was added and mQ water was added to get an end volume of 10 µl. In a separate tube 11 µl DNA Ligation Buffer was mixed carefully with 1 µl T4 Ligase until the solution was homogenous. This